- Email: [email protected]
Electrochemically produced divalent neodymium in chloride melt
Journal of Alloys and Compounds 408–412 (2006) 66–70
Electrochemically produced divalent neodymium in chloride melt Hajimu Yamana a,∗ , Byung Gi Park b , Osamu Shirai a , Toshiyuki Fujii a , Akihiro Uehara a , Hirotake Moriyama c a
c
Research Reactor Institute, Kyoto University, 2-1010, Asashiro Nishi, Kumatori-cho, Sennan-gun, Osaka 590-0494, Japan b FNC Technology Co. Ltd., SNU Research Innovation Cent., Gwanak-gu, Seoul 151-818, Republic of Korea Department of Nuclear Engineering, Graduate School of Kyoto University, Yoshida-honmachi, Sakyo-ku, Kyoto 606-8501, Japan Available online 8 June 2005
Abstract Production of Nd2+ by cathodic electrolysis of NdCl3 dissolved in molten LiCl–KCl was tested at 773 K. Cyclic voltammetry and differential pulse voltammetry were adopted for the determination of the formal redox potential between Nd3+ and its divalent or metallic states. UV–visible absorption spectrophotometry was carried out at the vicinity of the cathode during the simultaneous electrolysis at −1.9 V versus Ag/Ag+ . During the electrolysis, growth of a new broad absorption band ranging from 450 to 580 nm was observed and this was attributed to the absorption by the chloride complex of Nd2+ formed by the electrochemical reduction of NdCl3 . Black precipitate formed in the melt as a result of the constant potential electrolysis over long-period was analyzed by EPMA and ICP-AES and they were confirmed to be fine particles of Nd. The results were discussed in relation to the disproportionation of Nd2+ and reaction with quartz vessel. © 2005 Elsevier B.V. All rights reserved. Keywords: Neodymium; Divalent neodymium; Absorption spectrophotometry; Molten chloride; Voltammetry
1. Introduction The chemical characteristics of lanthanides in chloride melts merit further investigation for enhancing the chemical basis of lanthanides in molten salts and for developing techniques for lanthanide refining. One of the particular interests in relation to the performance of the electrolytic refining of lanthanides in molten salts is the formation of lanthanide’s less-common valence states [1–3]. Lanthanides having less energetic difference between electron configurations 4fn−1 6s2 5d1 and 4fn 6s2 5d0 are likely to form stable or meta-stable divalent state [4], especially in non-oxidizing conditions like molten chlorides. Nd is one of the lanthanides that have divalent compounds and it has been investigated in some particular organic solvents or solid matrix [5–7]. However, the formation of Nd2+ in LiCl–KCl eutectic melt has not been sufficiently investigated. The purpose of this study is to find the condition of the formation of Nd2+ by cathodic electrolysis of Nd3+ in molten LiCl–KCl ∗
Corresponding author. Tel.: +81 724 51 2442; fax: +81 724 51 2634. E-mail address: [email protected] (H. Yamana).
0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2005.04.104
and to examine the stability of produced Nd2+ in molten LiCl–KCl. A special focus is put on the disproportionation reaction and reaction with the vessel material of Nd2+ .
2. Experimental 2.1. Sample preparation All the experiments were carried out in a glove box system filled with dry Ar continuously purified to remove oxygen and humidity. The content of impurity O2 and H2 O in the inside atmosphere was continuously maintained less than 1 ppm. Anhydrous LiCl–KCl was purchased from APL Ltd. and used without further purification. Anhydrous NdCl3 purchased from APL Ltd. was used without further treatment for electrochemical analysis. Anhydrous NdCl3 , purchased from APL Ltd. and was used without further treatment for electrochemical analysis. Anhydrous NdCl3 purchased from Kojundo Chemical Laboratory Co. Ltd. was used for spectrophotometric measurements.
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
2.2. Cyclic voltammetry and differential pulse voltammetry 0.117 g of NdCl3 was dissolved into 11.178 g of molten LiCl–KCl at 773 K. Test tubes of quartz of 13 mm inner diameter were used for the measurement. A 1.0 mm Ø molybdenum rod was used for working electrode and a graphite rod of the same diameter was used for counter electrode, both covered with alumina sheath. A reference electrode, Ag/LiCl–KCl–AgCl (1 mol%) contained in a PYREX glass membrane tube was used. Polarization measurement system HZ-3000 (Hokuto-denko Co. Ltd.) was used for the voltammetry measurements. All measurements were performed at 773 ± 5 K. The differential pulse voltammetry was performed on a pulse period of 0.1 s with the square pulse having 0.01 V amplitude and 0.05 s width. 2.3. Absorption spectrophotometry under simultaneous electrolysis Absorption spectrophotometry under electrolysis was carried out using a transparent quartz cell illustrated in Fig. 1. The details of the absorption measurement system can be seen
67
elsewhere [8]. The tungsten working electrode was deeply immersed into the solution so that the light passed through the vicinity of its surface. The very thin tip of the PYREX glass at the end of the counter electrode worked as an electroconductive separator for the anode cell. Three kinds of working electrode were tested: tungsten mesh electrode (mesh made of 0.03 mm Ø tungsten wire, 50 mesh/in.); tungsten coil electrode made of 0.6 mm Ø tungsten wire; U-shaped electrode made of 0.8 mm Ø tungsten rod. These three electrodes were designed to give higher electrode surface area in the light path, but not to screen the light so much. The temperature of the sample during the experiment was controlled to 773 K within ±5◦ . 2.4. Long-period electrolysis followed by ICP-AES and EPMA analysis Using the same electrode arrangement to the voltammetry measurements, constant potential electrolysis of Nd3+ was performed using a quartz tube. The electrolysis was performed at −1.9 V versus Ag/Ag+ , which is the potential at which the reduction of Nd3+ to Nd2+ was expected to proceed. The melt was visually observed several times during the electrolysis and it was sampled after the end of the electrolysis, especially from the bottom of the cell. Qualitative element analysis was performed by EPMA for the frozen salt sample containing black precipitates, which had been formed during the electrolysis. The salt sample was once dissolved in water and the undissolved black precipitates were removed by filtration. The separated precipitates were completely dissolved in diluted hydrochloric acid and the amount of Nd was analyzed by ICP-AES. Powder X-ray diffraction pattern of the salt samples were taken by X-ray diffractometer.
3. Results and discussions 3.1. Voltammetry measurement
Fig. 1. Electrode arrangement for the spectrophotometry under simultaneous electrolysis.
Fig. 2 shows the cyclic voltammogram of NdCl3 in LiCl–KCl at 773. The potential was swept from natural potential, which is about −0.4 V versus Ag/Ag+ , toward negative direction and it had been reversed before the reduction of Li+ began. The reduction peak of Nd3+ → Nd0 is seen at about −2.1 V versus Ag/Ag+ and this is followed by an intensive oxidation peak. The peak potential of the major reduction current, which was taken at the scanning rate of 100 mV s−1 , was −2.062 V versus Ag/Ag+ for 773 K. A small growth of the reduction current can be seen in the region from −1.8 to −1.9 V versus Ag/Ag+ before the peak of Nd3+ → Nd0 grows, which can be assigned to the one-electron reduction of Nd3+ → Nd2+ . The oxidation peak of Fig. 2 appears to consist of two components and this is different from the sharper oxidation peaks shown by other lanthanides having no divalent states. Fig. 3 shows the result of the differential pulse voltammetry. At scanning rate of 20 mVs−1 , E1/2 of the small
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
68
Table 1 Results of the redox potential measurements E∗ (3–0) (V vs. Cl2 /Cl− ) E* (3–2) (V vs. Cl2 /Cl− ) E* (2–0) (V vs. Cl2 /Cl− ) ◦ Gdis (kJ)
Fig. 2. Cyclic voltammogram of Nd3+ in LiCl–KCl at 773 K.
peak and major peak were found to be −1.89 and −2.01 V versus Ag/Ag+ , respectively. It was found that the small peak did not depend on the potential scanning rate so much, but the major peak showed a slight dependence. Taking into consideration that these dependences on the scanning rate were not large, especially at lower scanning rates, they were treated as reversible reactions for estimating their formal potentials. Formal potentials of the reaction Nd3+ → Nd2+ was estimated from the E1/2 of differential pulse voltammetry using equation (1) [9], by adopting unity to the activity coefficient γ. That of Nd3+ → Nd0 was estimated from the Ep , peak potential of reduction current, of cyclic voltammetry using equation (2) [10], by adopting unity to the activity coefficient fox . RT D2+ 1/2 RT γ3+ E1/2 = Eo + ln + ln + Eref F D3+ F γ2+ (1)
Ep = Eo +
0.153 0.0568 log fox Co − + Eref , n n
−3.25 −3.178 −3.286 20.8
± ± ± ±
0.01 0.005 0.015 3.2
In these equations, D is a diffusion coefficient of suffixed species, Co the bulk concentration and Eref is the potential of reference electrode which is 1.263 V for Ag/LiCl–KCl–AgCl (1 mol%). Though D2+ and D3+ of Nd are unclear, since it is considered that the ratio of D2+ and D3+ of lanthanides does not vary largely because of their monotonic trend of the ionic radius contraction along the series, D2+ /D3+ of similar lanthanide was adopted as an approximation. In this paper, D2+ and D3+ of Eu, which have been reported [9], were adopted for Nd. These approximations resulted in larger errors of the determined potentials. Derived formal potentials, E∗ (3–2) and E∗ (3–0), are summarized in Table 1. E∗ (2–0) derived from other two was also shown in Table 1. 3.2. Spectrophotometry under simultaneous electrolysis Before starting the electrolysis, the absorption spectrum was obtained for the NdCl3 sample. The absorption spectrum of NdCl3 is shown in Fig. 4(a), in which an intensive hypersensitive absorption peak of Nd3+ can be seen at 589 nm. Electrolysis was performed at a potential about −1.9 V versus Ag/Ag+ by controlling the negative current to 1 or 2 mA. This
at 773 K (2)
Fig. 3. Differential pulse voltammogram of Nd3+ in LiCl–KCl at 773 K.
Fig. 4. UV–visible absorption spectrum of Nd3+ in LiCl–KCl under electrolysis.
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
potential was expected to allow the reaction Nd3+ → Nd2+ , but not to allow the formation of Nd0 by the reduction of Nd3+ . The absorption spectrum obtained about 13 min after the start of the electrolysis is shown in Fig. 4(b). Fig. 4(b) is a spectrum obtained by use of U-shaped rod electrode, but similar spectra were obtained for other two types of electrode. This spectrum is characterized by the appearance of a new broad absorption band in the region from 450 to 580 nm. As the time passed during the electrolysis, the spectrum gradually slanted. The inclination of the background line of the spectrum is possibly attributable to the increased light scattering or the slight change of the color of the silica window as a result of the electrolysis; it may be also due to a particular absorption by the product of the electrolysis. Fig. 4(c) indicates the spectrum obtained 30 min after the stop of the electrolysis. It should be emphasized that the broad absorption band in the region from 450 to 580 nm completely disappeared on the high inclined background within about 2 min after the electrolysis was stopped. The absorption peak at 589 nm observed before the electrolysis can be assigned to the hypersensitive f–f transitions of Nd3+ [11] and it did not change during the electrolysis. The broad absorption band ranging from 450 to 580 nm grew only during the electrolysis and it quickly disappeared after stopping the electrolysis. By the literatures about absorption spectra of divalent iodides of Nd in tetrahydroflan (THF) [5] and in CaF2 [7], it has been reported that Nd2+ in THF shows a broad absorption band ranging from 450 to 600 nm which consists of two peaks at 507 and 540 nm[5], and that Nd2+ in CaF2 [7] shows a broad absorption band over the region from 520 to 640 nm showing a maximum at 590. Since the broad absorption band was observed under a particular negative potential that corresponds to the reduction Nd3+ → Nd2+ and since its wave length region agrees well with that observed for NdI2 in THF and CaF2 , it should be considered that this broad absorption is of the electrochemically produced Nd2+ by the electrolysis in the melt. This assignment well agrees with the result by Terakado et al. [12]. In general, the electronic transitions, which give rise to the broad absorption bands of divalent lanthanides ion in various ligand fields are considered to be transitions between 4fn and 4fn−1 5d1 electron configurations [13]. Therefore, it should be a transition between 4f4 and 4f3 5d1 electron configurations of Nd2+ ion in LiCl–KCl. This should correspond to the transition from 5 I to 4 F d, which is about 20 kcm−1 in CaF2 matrix [13]. In general, f–d transition, which is Laporte allowed, shows very intensive molar absorptivity than f–f transition, but the observed absorbance is as high as hypersensitive peak of 589 nm. This means that the average concentration of formed Nd2+ was quite low, suggesting that quite limited amount of Nd2+ was observed only in the vicinity of the cathode. The large inclination of the base line, which is considered to be partly attributable to the change of the light scattering or window conditions, may also be attributable to the intervalence charge transfer [13].
69
3.3. Long-period electrolysis Electrolysis at −1.9 V versus Ag/Ag+ was performed for about 2 h using the same electrode arrangement as the spectrophotometry measurement. By the visual observation during the electrolysis, it was found that the melt entirely became black at the mid time, but that it became clear as the electrolysis proceeded. At the end of the electrolysis, very black precipitate settled down to the bottom of the cell with very clear melt left in the upper part. The inside surface of the quartz cell became lightly black and partly light brown. After the end of the electrolysis, the bottom of the quartz tube became completely black and lost its glassy nature. The qualitative chemical analysis of the black precipitate by EPMA and ICP-AES showed that the major component of this precipitate was Nd. On the other hand, the XRD pattern of the salt sample, which contained a little amount of fine black particles, did not obviously show the diffraction patterns of Nd metal or Nd oxide. 3.4. Stability of electrochemically produced Nd2+ The low observed absorbance and the rapid disappearance of this broad absorption within about 4 min after stopping the electrolysis, suggests the instability of the electrochemically produced Nd2+ in this system. The intensity of the broad absorption was approximately kept constant over the period later than about 10 min from the start of the electrolysis and this indicates that the generation and the disappearance of Nd2+ in the measured region were equilibrated during the electrolysis. According to the general understanding that the instability of divalent species with respect to its oxidation is in proportion to the instability with respect to the disproportionation [3], Nd2+ is likely to disproportionate into Nd0 and Nd3+ by reaction (3), 3Nd2+ (melt) → Nd0 + 2Nd3+ (melt)
(3)
where Nd3+ (melt) and Nd2+ (melt) means chloro-complex of these cations in the melt, NdCl6 3− and NdClx (x−2)− , respectively. The standard Gibbs energy change of reaction (3), ◦ Gdis , can be given by ◦
Gdis = 3F(E∗ (3–2) − E∗ (3–0)).
(4)
With the use of the formal potentials listed in Table 1, ◦ Gdis = 20.8 ± 3.2 kJ/mol is obtained for reaction (3) ◦ at 773 K. Due to this slightly positive value of Gdis , equilibrium (3) is likely to be sided to the left-side under the presence of the metal. However, under the continuous production of Nd2+ by the electrolysis, at the vicinity of the electrode surface, the local concentration of Nd2+ becomes high and leads to disproportionation into Nd0 and Nd3+ . Aggregation and settling down of the metallic fines accelerate the disproportionation more. Nd0 formed by reaction (3), as metallic fog, explains the dark color of the melt during the electrolysis and the settle down of the black precipitate to the
70
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
bottom. Clear XRD patterns of metal or oxide of Nd in the salt sample were not observed, but this suggests that the Nd precipitates were not in a clear crystallographic condition. The usage of the quartz cell must have accelerated the loss of produced Nd2+ by the reaction with SiO2 . Nd2+ (melt) is so unstable with respect to re-oxidation that it reacts with the quartz window to form oxide or oxichloride of Nd3+ and the color change of inside wall of the quartz tube to light black or brown should be the evidence of the reduction of SiO2 to SiO by Nd2+ . The bottom of the quartz cell encountered with large amount of black precipitates, of metallic Nd and it is considered to have reacted with Nd0 to form Nd oxide or oxichloride and broke the glass surface. 4. Conclusions
Acknowledgments Authors are indebted to Prof. Tsutomu Yamamura of Tohoku University for his technical advice in electrochemical experiments. The authors wish to thank Mr. Roy Jacobus for his help in improving the understandability of the English expressions of this paper.
References [1] [2] [3] [4] [5] [6]
By electrochemical measurements, the formal redox potentials of the reduction of Nd3+ to metal and Nd2+ in molten LiCl–KCl were determined. By means of UV–visible absorption spectrophotometry performed under simultaneous electrolysis of NdCl3 in molten LiCl–KCl, a broad absorption band ranging from 450 to 580 nm was observed and this was attributed to the Nd2+ formed by electrochemical reduction. Analysis and visual observation of the melt during the electrolysis showed the formation of Nd precipitates and this was attributed to the result of the disproportionation reaction of Nd2+ .
[7] [8] [9]
[10] [11] [12] [13]
A.N. Kamenskaya, Russ. J. Inorg. Chem. 29 (2) (1984) 251. L.B. Asprey, B.B. Cunningham, Prog. Inorg. Chem. 2 (1960) 267. D.A. Johnson, Adv. Inorg. Chem. Radiochem. 20 (1977) 1. L.J. Nugent, R.D. Baybarz, J.L. Burnett, J. Phys. Chem. 77 (12) (1973) 1528. A.N. Kamenskaya, N.B. Mikheev, N.P. Kholmogorova, Russ. J. Inorg. Chem. 28 (10) (1983) 1420. A.N. Kamenskaya, Dokl. Akad. Nauka SSSR 266 (2) (1982) 393 (in Russian). D.S. McClure, Z. Kiss, J. Chem. Phys. 39 (12) (1963) 3251. T. Nagai, T. Fujii, O. Shirai, H. Yamana, J. Nucl. Sci. Technol. 41 (6) (2004) 690. S.A. Kuznetsov, M. Gaune-Escard, Proceedings of the Workshop on Pyrochemical Separations, March 14–16, 2000, Avignon, France, 2000, pp. 295–305. T. Berzins, P. Delahay, J. Am. Chem. Soc. 75 (1953) 555. T. Fujii, H. Moiryama, H. Yamana, J. Alloys Compd. 351 (2003) L6. O. Terakado, S.Z. El Abedin, F. Endres, D. Nattland, W. Freyland, J. Non-Cryst. Solids 312–314 (2002) 459. K.E. Johnson, J.N. Sandoe, J. Chem. Soc. (A) (1969) 1695.
Electrochemically produced divalent neodymium in chloride melt Hajimu Yamana a,∗ , Byung Gi Park b , Osamu Shirai a , Toshiyuki Fujii a , Akihiro Uehara a , Hirotake Moriyama c a
c
Research Reactor Institute, Kyoto University, 2-1010, Asashiro Nishi, Kumatori-cho, Sennan-gun, Osaka 590-0494, Japan b FNC Technology Co. Ltd., SNU Research Innovation Cent., Gwanak-gu, Seoul 151-818, Republic of Korea Department of Nuclear Engineering, Graduate School of Kyoto University, Yoshida-honmachi, Sakyo-ku, Kyoto 606-8501, Japan Available online 8 June 2005
Abstract Production of Nd2+ by cathodic electrolysis of NdCl3 dissolved in molten LiCl–KCl was tested at 773 K. Cyclic voltammetry and differential pulse voltammetry were adopted for the determination of the formal redox potential between Nd3+ and its divalent or metallic states. UV–visible absorption spectrophotometry was carried out at the vicinity of the cathode during the simultaneous electrolysis at −1.9 V versus Ag/Ag+ . During the electrolysis, growth of a new broad absorption band ranging from 450 to 580 nm was observed and this was attributed to the absorption by the chloride complex of Nd2+ formed by the electrochemical reduction of NdCl3 . Black precipitate formed in the melt as a result of the constant potential electrolysis over long-period was analyzed by EPMA and ICP-AES and they were confirmed to be fine particles of Nd. The results were discussed in relation to the disproportionation of Nd2+ and reaction with quartz vessel. © 2005 Elsevier B.V. All rights reserved. Keywords: Neodymium; Divalent neodymium; Absorption spectrophotometry; Molten chloride; Voltammetry
1. Introduction The chemical characteristics of lanthanides in chloride melts merit further investigation for enhancing the chemical basis of lanthanides in molten salts and for developing techniques for lanthanide refining. One of the particular interests in relation to the performance of the electrolytic refining of lanthanides in molten salts is the formation of lanthanide’s less-common valence states [1–3]. Lanthanides having less energetic difference between electron configurations 4fn−1 6s2 5d1 and 4fn 6s2 5d0 are likely to form stable or meta-stable divalent state [4], especially in non-oxidizing conditions like molten chlorides. Nd is one of the lanthanides that have divalent compounds and it has been investigated in some particular organic solvents or solid matrix [5–7]. However, the formation of Nd2+ in LiCl–KCl eutectic melt has not been sufficiently investigated. The purpose of this study is to find the condition of the formation of Nd2+ by cathodic electrolysis of Nd3+ in molten LiCl–KCl ∗
Corresponding author. Tel.: +81 724 51 2442; fax: +81 724 51 2634. E-mail address: [email protected] (H. Yamana).
0925-8388/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2005.04.104
and to examine the stability of produced Nd2+ in molten LiCl–KCl. A special focus is put on the disproportionation reaction and reaction with the vessel material of Nd2+ .
2. Experimental 2.1. Sample preparation All the experiments were carried out in a glove box system filled with dry Ar continuously purified to remove oxygen and humidity. The content of impurity O2 and H2 O in the inside atmosphere was continuously maintained less than 1 ppm. Anhydrous LiCl–KCl was purchased from APL Ltd. and used without further purification. Anhydrous NdCl3 purchased from APL Ltd. was used without further treatment for electrochemical analysis. Anhydrous NdCl3 , purchased from APL Ltd. and was used without further treatment for electrochemical analysis. Anhydrous NdCl3 purchased from Kojundo Chemical Laboratory Co. Ltd. was used for spectrophotometric measurements.
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
2.2. Cyclic voltammetry and differential pulse voltammetry 0.117 g of NdCl3 was dissolved into 11.178 g of molten LiCl–KCl at 773 K. Test tubes of quartz of 13 mm inner diameter were used for the measurement. A 1.0 mm Ø molybdenum rod was used for working electrode and a graphite rod of the same diameter was used for counter electrode, both covered with alumina sheath. A reference electrode, Ag/LiCl–KCl–AgCl (1 mol%) contained in a PYREX glass membrane tube was used. Polarization measurement system HZ-3000 (Hokuto-denko Co. Ltd.) was used for the voltammetry measurements. All measurements were performed at 773 ± 5 K. The differential pulse voltammetry was performed on a pulse period of 0.1 s with the square pulse having 0.01 V amplitude and 0.05 s width. 2.3. Absorption spectrophotometry under simultaneous electrolysis Absorption spectrophotometry under electrolysis was carried out using a transparent quartz cell illustrated in Fig. 1. The details of the absorption measurement system can be seen
67
elsewhere [8]. The tungsten working electrode was deeply immersed into the solution so that the light passed through the vicinity of its surface. The very thin tip of the PYREX glass at the end of the counter electrode worked as an electroconductive separator for the anode cell. Three kinds of working electrode were tested: tungsten mesh electrode (mesh made of 0.03 mm Ø tungsten wire, 50 mesh/in.); tungsten coil electrode made of 0.6 mm Ø tungsten wire; U-shaped electrode made of 0.8 mm Ø tungsten rod. These three electrodes were designed to give higher electrode surface area in the light path, but not to screen the light so much. The temperature of the sample during the experiment was controlled to 773 K within ±5◦ . 2.4. Long-period electrolysis followed by ICP-AES and EPMA analysis Using the same electrode arrangement to the voltammetry measurements, constant potential electrolysis of Nd3+ was performed using a quartz tube. The electrolysis was performed at −1.9 V versus Ag/Ag+ , which is the potential at which the reduction of Nd3+ to Nd2+ was expected to proceed. The melt was visually observed several times during the electrolysis and it was sampled after the end of the electrolysis, especially from the bottom of the cell. Qualitative element analysis was performed by EPMA for the frozen salt sample containing black precipitates, which had been formed during the electrolysis. The salt sample was once dissolved in water and the undissolved black precipitates were removed by filtration. The separated precipitates were completely dissolved in diluted hydrochloric acid and the amount of Nd was analyzed by ICP-AES. Powder X-ray diffraction pattern of the salt samples were taken by X-ray diffractometer.
3. Results and discussions 3.1. Voltammetry measurement
Fig. 1. Electrode arrangement for the spectrophotometry under simultaneous electrolysis.
Fig. 2 shows the cyclic voltammogram of NdCl3 in LiCl–KCl at 773. The potential was swept from natural potential, which is about −0.4 V versus Ag/Ag+ , toward negative direction and it had been reversed before the reduction of Li+ began. The reduction peak of Nd3+ → Nd0 is seen at about −2.1 V versus Ag/Ag+ and this is followed by an intensive oxidation peak. The peak potential of the major reduction current, which was taken at the scanning rate of 100 mV s−1 , was −2.062 V versus Ag/Ag+ for 773 K. A small growth of the reduction current can be seen in the region from −1.8 to −1.9 V versus Ag/Ag+ before the peak of Nd3+ → Nd0 grows, which can be assigned to the one-electron reduction of Nd3+ → Nd2+ . The oxidation peak of Fig. 2 appears to consist of two components and this is different from the sharper oxidation peaks shown by other lanthanides having no divalent states. Fig. 3 shows the result of the differential pulse voltammetry. At scanning rate of 20 mVs−1 , E1/2 of the small
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
68
Table 1 Results of the redox potential measurements E∗ (3–0) (V vs. Cl2 /Cl− ) E* (3–2) (V vs. Cl2 /Cl− ) E* (2–0) (V vs. Cl2 /Cl− ) ◦ Gdis (kJ)
Fig. 2. Cyclic voltammogram of Nd3+ in LiCl–KCl at 773 K.
peak and major peak were found to be −1.89 and −2.01 V versus Ag/Ag+ , respectively. It was found that the small peak did not depend on the potential scanning rate so much, but the major peak showed a slight dependence. Taking into consideration that these dependences on the scanning rate were not large, especially at lower scanning rates, they were treated as reversible reactions for estimating their formal potentials. Formal potentials of the reaction Nd3+ → Nd2+ was estimated from the E1/2 of differential pulse voltammetry using equation (1) [9], by adopting unity to the activity coefficient γ. That of Nd3+ → Nd0 was estimated from the Ep , peak potential of reduction current, of cyclic voltammetry using equation (2) [10], by adopting unity to the activity coefficient fox . RT D2+ 1/2 RT γ3+ E1/2 = Eo + ln + ln + Eref F D3+ F γ2+ (1)
Ep = Eo +
0.153 0.0568 log fox Co − + Eref , n n
−3.25 −3.178 −3.286 20.8
± ± ± ±
0.01 0.005 0.015 3.2
In these equations, D is a diffusion coefficient of suffixed species, Co the bulk concentration and Eref is the potential of reference electrode which is 1.263 V for Ag/LiCl–KCl–AgCl (1 mol%). Though D2+ and D3+ of Nd are unclear, since it is considered that the ratio of D2+ and D3+ of lanthanides does not vary largely because of their monotonic trend of the ionic radius contraction along the series, D2+ /D3+ of similar lanthanide was adopted as an approximation. In this paper, D2+ and D3+ of Eu, which have been reported [9], were adopted for Nd. These approximations resulted in larger errors of the determined potentials. Derived formal potentials, E∗ (3–2) and E∗ (3–0), are summarized in Table 1. E∗ (2–0) derived from other two was also shown in Table 1. 3.2. Spectrophotometry under simultaneous electrolysis Before starting the electrolysis, the absorption spectrum was obtained for the NdCl3 sample. The absorption spectrum of NdCl3 is shown in Fig. 4(a), in which an intensive hypersensitive absorption peak of Nd3+ can be seen at 589 nm. Electrolysis was performed at a potential about −1.9 V versus Ag/Ag+ by controlling the negative current to 1 or 2 mA. This
at 773 K (2)
Fig. 3. Differential pulse voltammogram of Nd3+ in LiCl–KCl at 773 K.
Fig. 4. UV–visible absorption spectrum of Nd3+ in LiCl–KCl under electrolysis.
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
potential was expected to allow the reaction Nd3+ → Nd2+ , but not to allow the formation of Nd0 by the reduction of Nd3+ . The absorption spectrum obtained about 13 min after the start of the electrolysis is shown in Fig. 4(b). Fig. 4(b) is a spectrum obtained by use of U-shaped rod electrode, but similar spectra were obtained for other two types of electrode. This spectrum is characterized by the appearance of a new broad absorption band in the region from 450 to 580 nm. As the time passed during the electrolysis, the spectrum gradually slanted. The inclination of the background line of the spectrum is possibly attributable to the increased light scattering or the slight change of the color of the silica window as a result of the electrolysis; it may be also due to a particular absorption by the product of the electrolysis. Fig. 4(c) indicates the spectrum obtained 30 min after the stop of the electrolysis. It should be emphasized that the broad absorption band in the region from 450 to 580 nm completely disappeared on the high inclined background within about 2 min after the electrolysis was stopped. The absorption peak at 589 nm observed before the electrolysis can be assigned to the hypersensitive f–f transitions of Nd3+ [11] and it did not change during the electrolysis. The broad absorption band ranging from 450 to 580 nm grew only during the electrolysis and it quickly disappeared after stopping the electrolysis. By the literatures about absorption spectra of divalent iodides of Nd in tetrahydroflan (THF) [5] and in CaF2 [7], it has been reported that Nd2+ in THF shows a broad absorption band ranging from 450 to 600 nm which consists of two peaks at 507 and 540 nm[5], and that Nd2+ in CaF2 [7] shows a broad absorption band over the region from 520 to 640 nm showing a maximum at 590. Since the broad absorption band was observed under a particular negative potential that corresponds to the reduction Nd3+ → Nd2+ and since its wave length region agrees well with that observed for NdI2 in THF and CaF2 , it should be considered that this broad absorption is of the electrochemically produced Nd2+ by the electrolysis in the melt. This assignment well agrees with the result by Terakado et al. [12]. In general, the electronic transitions, which give rise to the broad absorption bands of divalent lanthanides ion in various ligand fields are considered to be transitions between 4fn and 4fn−1 5d1 electron configurations [13]. Therefore, it should be a transition between 4f4 and 4f3 5d1 electron configurations of Nd2+ ion in LiCl–KCl. This should correspond to the transition from 5 I to 4 F d, which is about 20 kcm−1 in CaF2 matrix [13]. In general, f–d transition, which is Laporte allowed, shows very intensive molar absorptivity than f–f transition, but the observed absorbance is as high as hypersensitive peak of 589 nm. This means that the average concentration of formed Nd2+ was quite low, suggesting that quite limited amount of Nd2+ was observed only in the vicinity of the cathode. The large inclination of the base line, which is considered to be partly attributable to the change of the light scattering or window conditions, may also be attributable to the intervalence charge transfer [13].
69
3.3. Long-period electrolysis Electrolysis at −1.9 V versus Ag/Ag+ was performed for about 2 h using the same electrode arrangement as the spectrophotometry measurement. By the visual observation during the electrolysis, it was found that the melt entirely became black at the mid time, but that it became clear as the electrolysis proceeded. At the end of the electrolysis, very black precipitate settled down to the bottom of the cell with very clear melt left in the upper part. The inside surface of the quartz cell became lightly black and partly light brown. After the end of the electrolysis, the bottom of the quartz tube became completely black and lost its glassy nature. The qualitative chemical analysis of the black precipitate by EPMA and ICP-AES showed that the major component of this precipitate was Nd. On the other hand, the XRD pattern of the salt sample, which contained a little amount of fine black particles, did not obviously show the diffraction patterns of Nd metal or Nd oxide. 3.4. Stability of electrochemically produced Nd2+ The low observed absorbance and the rapid disappearance of this broad absorption within about 4 min after stopping the electrolysis, suggests the instability of the electrochemically produced Nd2+ in this system. The intensity of the broad absorption was approximately kept constant over the period later than about 10 min from the start of the electrolysis and this indicates that the generation and the disappearance of Nd2+ in the measured region were equilibrated during the electrolysis. According to the general understanding that the instability of divalent species with respect to its oxidation is in proportion to the instability with respect to the disproportionation [3], Nd2+ is likely to disproportionate into Nd0 and Nd3+ by reaction (3), 3Nd2+ (melt) → Nd0 + 2Nd3+ (melt)
(3)
where Nd3+ (melt) and Nd2+ (melt) means chloro-complex of these cations in the melt, NdCl6 3− and NdClx (x−2)− , respectively. The standard Gibbs energy change of reaction (3), ◦ Gdis , can be given by ◦
Gdis = 3F(E∗ (3–2) − E∗ (3–0)).
(4)
With the use of the formal potentials listed in Table 1, ◦ Gdis = 20.8 ± 3.2 kJ/mol is obtained for reaction (3) ◦ at 773 K. Due to this slightly positive value of Gdis , equilibrium (3) is likely to be sided to the left-side under the presence of the metal. However, under the continuous production of Nd2+ by the electrolysis, at the vicinity of the electrode surface, the local concentration of Nd2+ becomes high and leads to disproportionation into Nd0 and Nd3+ . Aggregation and settling down of the metallic fines accelerate the disproportionation more. Nd0 formed by reaction (3), as metallic fog, explains the dark color of the melt during the electrolysis and the settle down of the black precipitate to the
70
H. Yamana et al. / Journal of Alloys and Compounds 408–412 (2006) 66–70
bottom. Clear XRD patterns of metal or oxide of Nd in the salt sample were not observed, but this suggests that the Nd precipitates were not in a clear crystallographic condition. The usage of the quartz cell must have accelerated the loss of produced Nd2+ by the reaction with SiO2 . Nd2+ (melt) is so unstable with respect to re-oxidation that it reacts with the quartz window to form oxide or oxichloride of Nd3+ and the color change of inside wall of the quartz tube to light black or brown should be the evidence of the reduction of SiO2 to SiO by Nd2+ . The bottom of the quartz cell encountered with large amount of black precipitates, of metallic Nd and it is considered to have reacted with Nd0 to form Nd oxide or oxichloride and broke the glass surface. 4. Conclusions
Acknowledgments Authors are indebted to Prof. Tsutomu Yamamura of Tohoku University for his technical advice in electrochemical experiments. The authors wish to thank Mr. Roy Jacobus for his help in improving the understandability of the English expressions of this paper.
References [1] [2] [3] [4] [5] [6]
By electrochemical measurements, the formal redox potentials of the reduction of Nd3+ to metal and Nd2+ in molten LiCl–KCl were determined. By means of UV–visible absorption spectrophotometry performed under simultaneous electrolysis of NdCl3 in molten LiCl–KCl, a broad absorption band ranging from 450 to 580 nm was observed and this was attributed to the Nd2+ formed by electrochemical reduction. Analysis and visual observation of the melt during the electrolysis showed the formation of Nd precipitates and this was attributed to the result of the disproportionation reaction of Nd2+ .
[7] [8] [9]
[10] [11] [12] [13]
A.N. Kamenskaya, Russ. J. Inorg. Chem. 29 (2) (1984) 251. L.B. Asprey, B.B. Cunningham, Prog. Inorg. Chem. 2 (1960) 267. D.A. Johnson, Adv. Inorg. Chem. Radiochem. 20 (1977) 1. L.J. Nugent, R.D. Baybarz, J.L. Burnett, J. Phys. Chem. 77 (12) (1973) 1528. A.N. Kamenskaya, N.B. Mikheev, N.P. Kholmogorova, Russ. J. Inorg. Chem. 28 (10) (1983) 1420. A.N. Kamenskaya, Dokl. Akad. Nauka SSSR 266 (2) (1982) 393 (in Russian). D.S. McClure, Z. Kiss, J. Chem. Phys. 39 (12) (1963) 3251. T. Nagai, T. Fujii, O. Shirai, H. Yamana, J. Nucl. Sci. Technol. 41 (6) (2004) 690. S.A. Kuznetsov, M. Gaune-Escard, Proceedings of the Workshop on Pyrochemical Separations, March 14–16, 2000, Avignon, France, 2000, pp. 295–305. T. Berzins, P. Delahay, J. Am. Chem. Soc. 75 (1953) 555. T. Fujii, H. Moiryama, H. Yamana, J. Alloys Compd. 351 (2003) L6. O. Terakado, S.Z. El Abedin, F. Endres, D. Nattland, W. Freyland, J. Non-Cryst. Solids 312–314 (2002) 459. K.E. Johnson, J.N. Sandoe, J. Chem. Soc. (A) (1969) 1695.